Catálogo de publicaciones - libros

This article concerns generating of 3D multiview exact models that are a complete representation of polyhedron, according to viewing sphere with perspective projection. Those models are going to be used for visual identification based on them and a scene depth map. We give a new concept and an algorithm for face-depended generation of multi-face views. It does not require any preprocessing nor auxiliary mechanisms or complex calculations connected with them

Two test images are decomposed into sequences of ten ordered images which result from a clustering of pixels. The first image is supposed to contain pixels belonging to , the tenth image — pixels belonging to . The remaining images gradually change from edge related to interior related. The clustering is provided by the so called Grade Correspondence — Cluster Analysis (GCCA), described in lastly published book on grade models and methods for data analysis. The GCCA is applied to the data matrices formed by a set of 12 variables which include gradient module, gray level, and ten variables describing the nearest neighborhood of each pixel according to the increasing level of module diffierentiation. Data matrices are visualized in form of the so called “ordered overrepresentation maps” and “grade stripcharts”.

In this paprer a method of automated detection of rectangular landmarks is presented. A landmark can be found if it has a monochromatic and characteristic color. The landmark doesn’t need to be visible as a whole, it can be partially obstructed by other objects.

This paper presents an efficient image registration technique using the Q-shift complex wavelet transform (Q-shift CWT). It is chosen for its key advantages compared to other wavelet transforms; such as shift invariance, directional selectivity, perfect reconstruction, limited redundancy and efficient computation. The experiments show that the proposed algorithm improves the computational efficiency and yields robust and consistent image registration compared with the classical wavelet transform.

The paper concerns the construction of the solids representation for a model-based visual identification system. The chosen representation is a complete 3D view model and the modelable solids are convex polyhedra and some class of non-convex polyhedra. A described algorithm uses the concepts of the view sphere with perspective projection and partitioning the sphere into the single-view areas.

A modified Hough transform has been proposed for grouping of image feature-carriers. The method has adjustable parameters, which are used in grouping and adding of missing image feature-carriers (due to registration noise). The adding of missing image features is based on performing of the secondary Hough-transform over the small window, in order to increase of transform resolution. The article contains of the research results addressing the influence of the method parameters on grouping of image features of real images.

Localization procedures for a mobile robot cooperating with external cameras and artificial navigation aids (landmarks) are discussed. An action planning method, taking into account in an exact way both the action cost and positioning uncertainty is presented. Its performance is illustrated by results of simulations.

The issues of spatial knowledge representation for mobile robots are considered. Two types of maps, grid and feature based, and two uncertainty representations, probabilistic and fuzzy are merged in one framework to obtain accurate and consistent geometric maps of the environment from range sensor readings.

In this paper the ellipse-specific contour grouping algorithm is introduced. It is a technique for detecting elliptical shapes from images. The algorithm uses ellipse-specific direct least square fitting and elliptic variance descriptor to assess an error of fit. A survey of other methods for detecting elliptical shapes is performed. Contour grouping techniques are discussed in details. The algorithm was illustrated on real images obtained from cytological phase contrast microscopy.

Active Shape Models (ASM) were proposed in the last decade of the 20 century as a versatile method of object localization and recognition. The theoretical concept on which the algorithm is based seems very attractive but the practical value of this technique still needs to be verified. The authors developed a multi-purpose object locating system containing an implementation of the ASMs and the experiments performed with the help of the system revealed serious draw-backs of the method. The discovered practical problems related to the use of the ASMs are presented in the paper.